Purpose To ascertain if a previous unpublished study–(Bratrud, 1998) in which physical self-perceptions and social physique anxiety substantially and independently predicted eating disorder risk in a convenience sample of female college athletes–could be cross validated. METHOD Fifty eight female college athletes from five sports completed the Physical Self-Perception Profile (PSPP), the Social Physique Anxiety Scale (SPAS), and the Drive For Thinness (THIN), Bulimia (BUL), and Body Dissatisfaction (BDIS) subscales of the Eating Disorder Inventory. They also completed two scales designed to measure socially desirable (SD) response tendencies (Paulhus, 1991). As in the initial study, and in line with extant literature (e.g., Williamson et al., 1994), it was assumed that BDIS mediates the risk and development of eating disorders. Thus, the data were analyzed to assess the unique contribution of BDIS to risk prediction (THIN and BUL) after SD, SPAS and PSPP had been statistically controlled for. RESULTS Analysis of the data from the cross validation study gave results that were conceptually and substantively similar to the earlier study in terms of the pattern of associations between variables, and in the levels of predictive contributions of PSPP, SPAS, and BDIS to eating disorder risk as measured by THIN. However, the results of the two studies were dissimilar with regard to prediction of risk as measured by BUL. Specifically, after controlling for SD, PSPP explained 25% of the variance in THIN, followed by substantial unique contributions from SPAS (15%) and BDIS (18%) respectively. In the overall hierarchical analyses, 62% of the variance in THIN was explained (compared to 66% previously), but only 20% of the variance in BUL was explained (compared to 31% previously). The degree of R2 shrinkage between studies suggests that the samples could be combined for an overall data analysis with respect to predicting THIN, but such a combination might be problematic for further study of the contribution of PSPP, and SPAS to the prediction of BUL (Pehazur, 1982). CONCLUSION These data add support to the hypothesis that PSPP and SPAS may make important and unique contributions to predicting eating disorder risk as measured by THIN, but the previous findings with respect to predictions of BUL were not substantively replicated. Overall, this attempt to cross validate the results of an earlier study was partially successful, and further study of the contributions of PSPP and SPAS to eating disorder risk seems to be warranted.